Re-engineered version of MINMOD: A clinical tool based on physiological modelling of the glucose-insulin system to quantitatively assess insulin sensitivity in healthy subjects and patients with type 2 diabetes mellitus

Abstract

Inden for sundhedsteknologi kan fysiologisk modellering inden for glukose-insulin-systemet være med til at øge forståelsen af diabetes, bl.a. gennem estimering af insulinsensitivitet (insulinfølsomhed) for raske personer og patienter med prædiabetes eller type 2-diabetes. Den gyldne standard inden for måling af insulinsensitivtet er den euglykæmiske hyperinsulinæmiske clamp-teknik, som er ret vanskelig og invasiv. Et alternativ til denne er Bergmans minimale model, som kan estimere insulinsensitivtet gennem fortolkning af den intravenøse glukosebelastningstest (IVGTT). Denne model har gennem en årrække vakt stor interesse i litteraturen hvortil, der endvidere er lavet en PC-applikation til (MINMOD Millennium). Denne er desværre begrænset til en enkelt optimeringsalgoritme næsten uden nogen form for brugertilpasning udover også at virke forældet. Derfor bestræbede dette kandidatspeciale sig på at rekonstruere en version af MINMOD med fokus på mere brugertilpasning af optimeringsprocessen for bl.a. at kunne estimere insulinsensitivitet. En tilgang med ikke-lineær mindste kvadraters metode (NLS) blev benyttet sammen med Bergmans minimale model til at udvikle en PC-applikation i MATLAB (version R2015a). Relevant IVGTT-data fra litteraturen blev fundet, ekstraheret, processeret samt implementeret i den rekonstruerede version af MINMOD. Alt dette blev samlet og præsenteret i en brugergrænseflade (GUI), mens softwareudviklingsprocessen hovedsageligt var baseret på vandfaldsmodellen samt usability-heuristikker. Under softwareudviklingsprocessen var det muligt at rekonstuere en version af MINMOD i et æstetisk og minimalistisk GUI. Det var her muligt at tilpasse i optimeringsprocessen mht. estimeringen af relevante parametre, herunder insulinsensitivitet, inden for Bergmans minimale model. Overordnet var det muligt at bekræfte estimeringen af parametrene med andre studier fundet i litteraturen, der validerer den rekonstuerede version af MINMOD til et vist omfang. For at opnå en højere valideringsgrad af programmet, er det dog nødvendigt med flere tests.

In the field of biomedical engineering, physiological modelling of the glucose-insulin system may improve understanding of diabetes by e.g. estimating the insulin sensitivity in healthy individuals and patients with prediabetes or type 2 diabetes. The gold standard of measuring insulin sensitivity is the euglyceamic hyperinsulinemic clamp technique, which is rather tedious, invasive, and labour-intensive. A comparable alternative to this is the Bergman minimal model, assessing insulin sensitivity by the intravenous glucose tolerance test (IVGTT). This model has gained much attention in the literature and has been implemented as the PC application MINMOD Millennium, which unfortunately is limited to one optimisation algorithm almost without any level of customisation besides being rather outdated. Therefore, this MSc Thesis aimed to re-engineer a version of MINMOD, with focus on providing more customisation of the optimisation process to estimate e.g. insulin sensitivity. A nonlinear least squares (NLS) approach was used together with the Bergman minimal model to develop a PC application in MATLAB (version R2015a). Relevant IVGTT data from the literature was found, extracted, prepared, and implemented into the re-engineered version of MINMOD. This was presented in a graphical user interface (GUI), and the software development process was mainly based on the waterfall model and usability heuristics. During the software development process, it was possible to re-engineer a version of MINMOD in an aesthetic and minimalistic GUI. Here, it was possible to customise the optimisation process regarding estimates of relevant parameters, including insulin sensitivity, of the Bergman minimal model. Overall, parameter estimates could be replicated and compared with other studies found in literature, validating the re-engineered version of MINMOD to some degree. However, more testing is required in order to obtain an even higher degree of validation.

A master thesis from Aalborg University

Re-engineered version of MINMOD: A clinical tool based on physiological modelling of the glucose-insulin system to quantitatively assess insulin sensitivity in healthy subjects and patients with type 2 diabetes mellitus

[Rekonstrueret version af MINMOD: Et klinisk værktøj baseret på fysiologisk modellering af glukose-insulin-systemet til kvantitativ estimering af insulinsensitivitet ved raske personer og patienter med type 2-diabetes mellitus]